cellmap_flow.finetune.finetune_cli

Command-line interface for LoRA finetuning.

Usage:

python -m cellmap_flow.finetune.finetune_cli –model-checkpoint /path/to/checkpoint –corrections corrections.zarr –output-dir output/fly_organelles_v1.1

# With custom settings python -m cellmap_flow.finetune.finetune_cli –model-checkpoint /path/to/checkpoint –corrections corrections.zarr –output-dir output/fly_organelles_v1.1 –lora-r 16 –batch-size 4 –num-epochs 20 –learning-rate 2e-4

Attributes

logger

Classes

RestartController

In-memory restart control shared between training loop and server endpoint.

Functions

build_arg_parser()
main()

Module Contents

cellmap_flow.finetune.finetune_cli.logger

class cellmap_flow.finetune.finetune_cli.RestartController

In-memory restart control shared between training loop and server endpoint.

request_restart(payload: dict | None) → bool

get_if_triggered() → dict | None

cellmap_flow.finetune.finetune_cli.build_arg_parser()

cellmap_flow.finetune.finetune_cli.main()